LETTER

MX2 is an interferon-induced inhibitor of HIV-1 infection

doi:10.1038/nature12653

Melissa Kane1,2, Shalini S. Yadav1,2,3{, Julia Bitzegeio1,2, Sebla B. Kutluay1,2, Trinity Zang1,2,3, Sam J. Wilson1,2,3{, John W. Schoggins4{, Charles M. Rice4, Masahiro Yamashita1, Theodora Hatziioannou1 & Paul D. Bieniasz1,2,3

HIV-1 replication can be inhibited by type I interferon (IFN), and the expression of a number of gene products with anti-HIV-1 activity is induced by type I IFN1,2. However, none of the known antiretroviral proteins can account for the ability of type I IFN to inhibit early, preintegration phases of the HIV-1 replication cycle in human cells3,4. Here, by comparing gene expression profiles in cell lines that differ in their ability to support the inhibitory action of IFN-a at early steps of the HIV-1 replication cycle, we identify myxovirus resistance 2 (MX2) as an interferon-induced inhibitor of HIV-1 infection. Expression of MX2 reduces permissiveness to a variety of lentiviruses, whereas depletion of MX2 using RNA interference reduces the anti-HIV-1 potency of IFN-a. HIV-1 reverse transcription proceeds normally in MX2-expressing cells, but 2-long terminal repeat circular forms of HIV-1 DNA are less abundant, suggesting that MX2 inhibits HIV-1 nuclear import, or destabilizes nuclear HIV-1 DNA. Consistent with this notion, mutations in the HIV-1 capsid protein that are known, or suspected, to alter the nuclear import pathways used by HIV-1 confer resistance to MX2, whereas preventing cell division increases MX2 potency. Overall, these findings indicate that MX2 is an effector of the anti-HIV-1 activity of type-I IFN, and suggest that MX2 inhibits HIV-1 infection by inhibiting capsid-dependent nuclear import of subviral complexes. We and others have previously identified proteins with antiretroviral activity on the basis of their differential expression in cells that are permissive or non-permissive with respect to particular steps in the HIV-1 life cycle5,6. We noticed that monocytoid cell lines varied in their ability to support the anti-HIV-1 activity of type I IFN. Specifically, IFN-a treatment of THP-1 cells caused an ,40-fold reduction in infection by an HIV-1-based green fluorescent protein (GFP) reporter vector, whereas treatment of K562 and U937 cells had little effect (Fig. 1a). When these cell lines were differentiated into a macrophagelike state by treatment with phorbol 12-myristate 13-acetate (PMA), the inhibitory effect of IFN-a was accentuated in THP-1 cells and accentuated to a lesser extent in U937 cells, but remained nearly absent in K562 cells (Fig. 1a). To identify candidate effectors of the antiviral action of IFN-a, we used microarrays to measure messenger RNA levels in the aforementioned cell lines. Twenty-two genes whose induction, or non-induction, by IFN-a correlated to varying degrees with the ability or inability of IFN-a to inhibit HIV-1GFP vector infection in the monocytoid cell lines were selected for further study (Fig. 1b and Extended Data Figs 1 and 2). Among these candidates, MX2, a gene that was not previously thought to exhibit antiviral activity7, was of particular interest as we recently identified it as a hit in an overexpression screen in a T-cell line during which MX2 modestly inhibited infection by HIV-1 (ref. 8). Western blot analyses confirmed that MX2 expression was strongly induced by IFN-a in THP-1 cells but not K562 cells, and a basal level of MX2 expression was slightly increased by IFN-a treatment in U937 cells
1

(Fig. 1c). MX2 was expressed at a basal level in primary CD41 T cells and macrophages, and was induced to varying degrees by IFN-a, depending on the individual donor, and how cells were activated (Extended Data Fig. 3). Expression of the 22 candidate and control genes in K562 cells revealed that only MX2 and a control antiviral gene coding for rhesus macaque TRIM5-a9 inhibited HIV-1 infection. (Fig. 2a). A rhesus macaque variant of MX2 also inhibited HIV-1 infection to a similar degree as human MX2, whereas MX1 was inactive against HIV-1 (Fig. 2a), even though it inhibits a variety of other viruses7. Although MX2 clearly inhibited HIV-1 infection (Fig. 2ad), the fact that U937 cells (Fig. 1a), primary macrophages and anti-CD3/CD28-stimulated CD41 T cells are readily infected by HIV-1, despite expressing appreciable levels of MX2 (Fig. 1c and Extended Data Fig. 3), indicates that the block imposed by MX2 is not absolute, or that MX2 potency is perhaps influenced by the cellular environment or cofactors. MX1 and MX2 are members of a family of dynamin-like GTPases7, but only MX2 is localized to the nucleus by virtue of a basic nuclear localization signal (NLS) contained within its amino-terminal 25 amino acids10,11. Notably, the N-terminal 25 amino acids that encode the MX2 NLS were strictly required for antiviral activity (Fig. 2b, c). Conversely, the mutations K131A and T151Awhich inhibit GTP binding and hydrolysis, respectively11did not block the anti-HIV-1 activity of MX2 (Fig. 2b, c). This result is in contrast to findings with MX1, whose antiviral activity is GTPase dependent7, but should be interpreted cautiously given the reported ability of these MX2 mutants to induce a generalized perturbation of nucleocytoplasmic transport11. In addition to its activity against HIV-1 and HIV-2 (Fig. 2d), MX2 expression in HOS cells inhibited infection by GFP reporter viruses based on a variety of primate lentiviruses, including simian immunodeficiency viruses SIVMAC, SIVAGMTan and SIVAGMSab, with some variation in MX2 antiviral potency (Fig. 2e). The nonprimate lentivirusesequine infectious anaemia virus and feline immunodeficiency viruswere less potently inhibited, whereas a gammaretrovirusmurine leukaemia viruswas only marginally sensitive to MX2. The experiments described above all represented single-cycle infection assays, using vesicular stomatitis virus glycoprotein (VSV-G) -pseudotyped reporter viruses. However, expression of MX2 in GHOSTR5 cells also inhibited infection by two full-length primary HIV-1 strains, suggesting that MX2 inhibition was independent of the route of entry, and not counteracted by HIV-1 accessory genes (Fig. 3a). Moreover, MX2 expression in GHOST-X4 cells inhibited spreading infection by full-length replication-competent HIV-1NL4-3 (Fig. 3b), reducing the number of infected cells by ,20-fold during the exponential phase of viral growth. Reduction of MX2 expression in THP-1 cells (Fig. 3c, d) or in HOS cells (Extended Data Fig. 4a, b) using short hairpin RNAs (shRNAs) reduced, but did not eliminate, the antiviral effect of IFN-a. Thus, MX2 is required for the full potency of IFN-a,

Aaron Diamond AIDS Research Center, New York, New York 10016, USA. 2Laboratory of Retrovirology, The Rockefeller University, New York, New York 10065, USA. 3Howard Hughes Medical Institute, New York, New York 10016, USA. 4Center for the Study of Hepatitis C, The Rockefeller University, New York, New York 10065, USA. {Present addresses: University of Texas Southwestern Medical Center, 5323 Harry Hines Boulevard, Dallas, Texas 75390-9048, USA (J.W.S.); MRC Centre for Virus Research, Institute of Infection, Immunity and Inflammation, College of Medical, Veterinary and Life Sciences, University of Glasgow, Glasgow G12 8QQ, UK (S.J.W.); Weill Cornell Medical College, 525 East 68th street, New York, New York 10025, USA (S.S.Y.). 2 4 O C T O B E R 2 0 1 3 | VO L 5 0 2 | N AT U R E | 5 6 3

2013 Macmillan Publishers Limited. All rights reserved

RESEARCH LETTER
a
GFP positive (%)
100 10 1 0.1 0.1 1 10 100 No IFN THP-1 + IFN 100 10 1 0.1 0.1 K562 100 (PMA) 10 1 0.1 0.1 1 10 100 0.1 1 10 100 1 10 100 K562 100 10 1 0.1 0.1 U937 100 (PMA) 10 1 10 100 U937

a
Infected (GFP+) cells (%)
100

m.o.i. = 0.6

m.o.i. = 0.2

m.o.i. = 0.07

10

GFP positive (%)

THP-1 100 (PMA) 10 1 0.1

delN25

K131A

Infected (GFP+) cells (%)

100 80 60 40 20 0 0 1 2

Titre (IU ml1)

1 0.1 0.1 1 10 100

MX2-

Vector

T151A

MX2

Inoculum (l)

Inoculum (l)

Inoculum (l)
Tubulin-

to r M X2

Array signal in IFN--treated cells (a.u.)

MX2 5,000

MX2

Virus inoculum (m.o.i.)

500 THP-1 50 50,000 500 5,000 50,000 MX2 MX2 50 MX2 K562 50 U937 500 5,000 50,000

e
2.0107

SIVMAC

SIVAGMTan

SIVAGMSab

EIAV

Ve c

FIV

Ve c
0.0
Ve

1.5108 1.0108 5.0107

0.0
o M r X2

5105 4105 3105 2105 1105

0

1.5104 1.0104 5.0103

0.0
ct o M r X2 Ve

1106 8105 6105 4105 2105

0
ct o M r X2 Ve

1.5106 1.0106 5.0105

Titre (IU ml1)

50

1.5107 1.0107 0.5107

0.0
ct Ve

500 5,000 50,000

MX2

ct o M r X2

X2

ct

ct

500 THP-1 (PMA) 50 500 5,000 50,000 50 500 K562 (PMA) 5,000 50,000 50 U937 (PMA) 500 5,000 50,000

50

Array signal in untreated cells (a.u.)

c
IFN- (U ml1) MX2Tubulin1 0

THP-1 1,000 100 10

K562 1,000 100 10

U937 1,000 100 10

Figure 1 | Differential effects of IFN-a on HIV-1 infection of monocytoid cell lines correlates with MX2 expression. a, Undifferentiated (top) or PMA-differentiated (bottom) THP-1, K562 and U937 cells with or without IFN-a treatment (1,000 U ml21) were challenged with a GFP-expressing HIV-1 vector (CSGW). b, RNA extracted from cells treated identically to those shown in a was analysed on microarrays. The array signal is plotted in arbitrary units (a.u.), and the data points representing MX2 are highlighted. c, Western blot analysis of MX2 and tubulin expression in monocytoid cell lines treated for 24 h with the indicated doses of IFN-a. Numbers below each lane indicated fold increase in MX2 protein levels relative to untreated cells. ND, not detected.

Figure 2 | Inhibition of lentivirus infection by wild-type and mutant MX2, but not other differentially interferon-induced genes. a, Infection of K562 cells, previously transduced with an HIV-1 vector (SCRPSY) expressing negative (luciferase) or positive (rhesus macaque (rh)TRIM5-a-coding) control genes, or candidate antiviral genes, with GFP-expressing HIV-1 vector (CSGW) at the indicated multiplicity of infection (m.o.i.). b, Western blot analysis of MX2 and tubulin expression in K562 cell clones transduced with an HIV-1 vector (CSIB) expressing wild-type and mutant MX2 proteins. delN25, MX2 mutant lacking the N-terminal NLS. c, Infection of the same K562 cells as in b with an HIV-1GFP reporter virus. d, e, Infection of HOS cells, previously transduced with an MX2-expressing or empty HIV-1 vector (SCRPSY), with various GFP reporter viruses. Titres are mean 1 s.d., n 5 3 technical replicates, representative of four experiments. EIAV, equine infectious anaemia virus; FIV, feline immunodeficiency virus; MLV, murine leukaemia virus.

14

38

45

1.6

2.4

ND

ND

ND

ND

2.9

but is not solely responsible for the inhibitory action of IFN-a on the early steps of the HIV-1 replication cycle. Consistent with this conclusion, IFN-a treatment reduced the accumulation of HIV-1 reverse transcripts in HOS cells (Fig. 4a), as has previously been reported for other cell types12. Conversely, MX2 expression did not inhibit reverse transcript accumulation in either HOS or K562 cells (Fig. 4a and Extended Data Fig. 5). However, MX2 did reduce the generation of 2-long terminal repeat (2-LTR) circles (Fig. 4a and Extended Data Fig. 5), which are thought to form only after retroviral DNA has accessed the nucleus of infected cells. MX2 may, therefore, inhibit the entry of HIV-1 into the nucleus, or perhaps cause destabilization of viral DNA in the nucleus. Consistent with previous
5 6 4 | N AT U R E | VO L 5 0 2 | 2 4 O C T O B E R 2 0 1 3

reports10,11, we found that that N- or carboxy-terminally haemagglutinintagged forms of MX2 were particularly concentrated at nuclear pores marked by the nucleoporin NUP98 (Extended Data Fig. 6). The MX2(K131A) mutant is primarily cytoplasmic but nevertheless inhibits nucleocytoplasmic transport11 and also retains antiviral activity (Fig. 2c). Therefore, alteration of the fate of incoming HIV-1 DNA with respect to the nucleus may underlie the antiviral activity of MX2, even though stable physical association with nuclear pores may not be required for antiviral function. The HIV-1 capsid protein (CA) is a key determinant required for infection of non-dividing cells and nuclear entry of subviral complexes1315. Indeed, HIV-1 CA mutations have been shown to change the requirement for specific nucleoporins (for example, NUP358 (also known as RANBP2), NUP85, NUP153, NUP155) during HIV-1 infection, and to alter the distribution of sites at which HIV-1 DNA integrates into host chromosomes1618. Therefore, we tested whether a number of CA mutations that are known or suspected to affect the pathway used by HIV-1 DNA into the nucleus also affected sensitivity to inhibition by MX2 (Fig. 4b). Of these, a mutation (N57S) that confers cell cycle dependence on HIV-1 infection19, and presumably restricts HIV-1 nuclear entry to the mitotic phase of the cell cycle, conferred resistance

2013 Macmillan Publishers Limited. All rights reserved

Ve

Ve

X2

or

or

5,000

to r M X2
MLV

50,000

N Lu Ve one rh cife cto T R ra r IM se 5 G BP G 1 H BP ER 5 C IF 6 I PA M44 RP X2 SI R 10 G GL LE 1 SP C1 1 TA00 TNP1 K2 EPETV ST 7 H I1 H CP ER 5 H C SP 5 A LO IF 6 C IF I35 40 IH 0 RB7591 SE R M4 RPSA 3 IN D2 G 1 rh M X2 M X1

Vector MX2

T151A K131A delN25

d
6107 4107 2107 0

HIV-1 8107 6107 4107 2107 0

HIV-2

LETTER RESEARCH
Infected (GFP+) cells (%)

a
Titre (IU ml1)

HIV-1TF2864 1.5105 1.0105 5.0104 0.0 1.5105 1.0105 5.0104 0.0

Ve

HIV-1TF2851

100 Vector MX2 MX2(T151A)

a
RT products (GFP, copies per l)

40,000 30,000 20,000 10,000 0 0 5 10 15 20 25

150

2-LTR circles (copies per l)

100 50 0

10

None IFN- MX2 Nevirapine

5 10 15 20 25

Ve ct or M X2

ct or M X2

Days after infection

Time after infection (h)

Time after infection (h)

G89V 1.2107 8.0106 4.0106 0.0
to M r X2
to M r X2 Ve c

Fold titre increase in shRNA(MX2) cells

c
IFN- U ml1 MX2-

shRNA(ctrl) shRNA(MX2)
0 10 100 1,000 0 10 100 1,000

d
107

7 17

b
5
3107

WT 1.2106 8.0105 4.0105 0.0

to M r X2

N57S 1.5106 1.0106 5.0105 0.0

to M r X2

A92E 1.5107 1.0107 5.0106 0.0

G94D 3107 2107 1107 0

N74D

4 3 2 1

Titre (IU ml1)

Titre (IU ml1)

2107 1107 0
Ve c

106

105

Ve c

Ve c

to M r X2

0.2 2.1 5.6

6.6

0.1 0.2 0.4

1.7

104

1, 00 No 0 U IFN m - l 1 IF N

c
Fold inhibition

R sh NA RN (c A( trl) M X2 sh ) RN sh A RN (c t A( rl) M X2 )

25 20 15 10 5 0 106

Infected (GFP+) cells (%)

HIV-1 (WT) HIV-1 (N57S)

d
100 10 1

Dividing 100 10 1 Vector MX2 0.1 0.01 0.10 1 0.1 0.01

Ve c

Non-dividing

P = 0.27

sh

P = 0.001

P = 0.004

Figure 3 | MX2 inhibits replication-competent HIV-1 and is required for the full antiviral activity of IFN-a. a, Infection of empty vector (CSIB) or MX2-expressing GHOST-R5 cells with full-length primary HIV-1 strains. Titres are mean 1 s.d., n 5 3 technical replicates, representative of two experiments. b, Growth of replication-competent HIV-1NL4-3 in empty vector (CSIB) or MX2-expressing GHOST-X4 cells (containing an HIV-2-LTR-GFP gene). c, Western blot analysis of MX2 and tubulin expression in IFN-a-treated THP-1 cells expressing control or MX2-targeted shRNAs. Numbers below each lane indicate fluorescence intensity associated with the MX2 band. d, HIV-1 GFP reporter virus infection of shRNA-expressing THP-1 cells from c, with (black) or without (white) IFN-a treatment. Titres are mean 1 s.d., n 5 3 technical replicates, P values calculated using unpaired t-test, representative of three experiments.

Titre (IU ml1)

105 104 103

0.10

Ve c

to M r X2

Tubulin-

Virus inoculum (l) Virus inoculum (l)

to MX2 (Fig. 4b). Another mutation, G89V, which abolishes cyclophilin A binding by HIV-1 CA and the requirement for NUP358 during HIV-1 infection17, also conferred apparently complete MX2 resistance (Fig. 4b). Another CA mutation, N74D, which abolishes CA interaction with cleavage and polyadenylation specificity factor 6 (CPSF6)16, reduced but did not eliminate sensitivity to MX2, whereas the mutations G94D and A92E, which confer cyclophilin A sensitivity (cyclosporin A dependence) during early replication steps20, slightly reduced MX2 sensitivity (Fig. 4b). These data demonstrate that the viral capsid governs the sensitivity of HIV-1 to MX2. In addition, they show that the antiviral activity of MX2 is specific, and unlikely to be the result of some generalized perturbation of cell physiology. Notably, the MX2resistant CA mutant N57S exhibited a modest degree of resistance to IFN-a, relative to wild-type HIV-1, in THP-1 cells and HOS cells (Fig. 4c and Extended Data Fig. 7), supporting the notion that MX2 is one, but not the only, effector of the antiviral activity of IFN-a during the early steps of the HIV-1 replication cycle. Because the cell-cycle-dependent HIV-1 CA mutant N57S was not inhibited by MX2 (Fig. 4b), we reasoned that arresting the cell cycle and thereby restricting HIV-1 infection to non-mitotic cells might potentiate the antiviral activity of MX2. Growth arrest of HOS or K562 cells with aphidicolin blocked infection by a control cell-cycledependent retrovirus (murine leukaemia virus) irrespective of MX2 expression, whereas HIV-1 was almost unaffected, as expected (Fig. 4d and Extended Data Fig. 8a, b). However, the inhibitory activity of MX2 was increased in non-dividing cells (Fig. 4d and Extended Data Fig. 8), in which it inhibited a single cycle of replication by ,30-fold. In other words, MX2 both inhibited and conferred a degree of cell cycle dependence on wild-type HIV-1 infection. Type I IFN inhibits HIV-1 replication at multiple points in the life cycle, both before and after the point at which MX2 seems to act2,6,12.

Figure 4 | MX2 activity reduces levels of nuclear HIV-1 DNA, is capsid dependent and is more potent in non-dividing cells. a, Quantitative PCR analysis of reverse transcript (RT, left) and 2-LTR circle (right) abundance in inhibitor-treated or MX2-expressing HOS cells. b, Wild-type (WT) or CA-mutant HIV-1GFP reporter virus infection of vector or MX2-expressing HOS cells. Titres are mean 1 s.d., n 5 3 technical replicates, representative of four experiments. c, Infectivity of wild-type and N57S CA-mutant HIV-1GFP reporter viruses in untreated and IFN-a-treated THP-1 cells. Titres are mean 1 s.d., n 5 3 technical replicates, representative of four experiments. Fold inhibition is the ratio of the mean titres on untreated and IFN-a-treated cells. d, HIV-1GFP reporter virus infection of dividing and non-dividing (aphidicolin-treated) vector- or MX2-expressing HOS cell clones.

Thus MX2 is one of multiple effectors that contribute to the overall anti-HIV-1 activity of type I IFN. A few potential mechanisms might underlie the anti-HIV-1 activity of MX2. First, MX2 might directly target the incoming viral capsid, in a manner akin to the primate TRIM5-a and murine Fv1 antiretroviral proteins2,9,21, or mutant cytoplasmic forms of CPSF6 (ref. 16). As with MX2, one consequence of the action of these capsid-targeting proteins is inhibition of the import of viral DNA into the nucleus, and in some cases their potency is enhanced in non-dividing cells16,22. A second possibility is that MX2 inhibits particular nuclear import pathways, without regard to the precise nature of the import cargo, as mutant forms of MX2 have been shown to inhibit the nuclear accumulation of model cargos unrelated to HIV-1 (ref. 11). A third possibility is that MX2 acts after nuclear entry to destabilize viral DNA and/or inhibit integration. In these scenarios, CA mutations (G89V, N57S) could confer resistance by inhibiting interaction with MX2, by modulating the timing or extent of capsid uncoating, or by directing HIV-1 to alternative nuclear entry pathways. We note that the MX2-resistant G89V and N57S mutants exhibit reduced infectiousness in human cells, raising the possibility that the mutations abolish the use of pathways or processes during infection that are inhibited by MX2. Finally, it is possible that MX2 acts indirectly, for example by affecting the nuclearcytoplasmic distribution of other cellular proteins that can interact with the viral capsid. However, the poor correlation in the degree of MX2 (Fig. 4) and CPSF6
2 4 O C T O B E R 2 0 1 3 | VO L 5 0 2 | N AT U R E | 5 6 5

2013 Macmillan Publishers Limited. All rights reserved

N 10 10 1 0 10,000 ,0 0 N 0 o IF N 10 10 1 0 10,000 ,0 00

IF

RESEARCH LETTER
(ref. 23) resistance/sensitivity exhibited by HIV-1 CA mutants suggests that redistribution of CPSF6 is unlikely to underlie the antiviral action of MX2. Although further work will be required to precisely define the molecular mechanisms involved, our findings demonstrate that MX2 is an effector in the anti-HIV-1 activity of type I IFN and underscore the remarkable diversity of proteins that cells can mobilize as antiretroviral defences.
6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. Online Content Any additional Methods, Extended Data display items and Source Data are available in the online version of the paper; references unique to these sections appear only in the online paper. Received 23 August; accepted 12 September 2013. Published online 13 October 2013. 1. 2. 3. Ho, D. D. et al. Recombinant human interferon alfa-A suppresses HTLV-III replication in vitro. Lancet 325, 602604 (1985). Neil, S. & Bieniasz, P. Human immunodeficiency virus, restriction factors, and interferon. J. Interferon Cytokine Res. 29, 569580 (2009). Bitzegeio, J., Sampias, M., Bieniasz, P. D. & Hatziioannou, T. Adaptation to the interferon-induced antiviral state by human and simian immunodeficiency viruses. J. Virol. 87, 35493560 (2013). Goujon, C. et al. Evidence for IFNa-induced, SAMHD1-independent inhibitors of early HIV-1 infection. Retrovirology 10, 23 (2013). Sheehy, A. M., Gaddis, N. C., Choi, J. D. & Malim, M. H. Isolation of a human gene that inhibits HIV-1 infection and is suppressed by the viral Vif protein. Nature 418, 646650 (2002). Neil, S. J., Zang, T. & Bieniasz, P. D. Tetherin inhibits retrovirus release and is antagonized by HIV-1 Vpu. Nature 451, 425430 (2008). Haller, O., Staeheli, P. & Kochs, G. Interferon-induced Mx proteins in antiviral host defense. Biochimie 89, 812818 (2007). Schoggins, J. W. et al. A diverse range of gene products are effectors of the type I interferon antiviral response. Nature 472, 481485 (2011). Stremlau, M. et al. The cytoplasmic body component TRIM5a restricts HIV-1 infection in Old World monkeys. Nature 427, 848853 (2004). n, K. et al. Human MxB protein, an interferon-a-inducible GTPase, contains a Mele nuclear targeting signal and is localized in the heterochromatin region beneath the nuclear envelope. J. Biol. Chem. 271, 2347823486 (1996). King, M. C., Raposo, G. & Lemmon, M. A. Inhibition of nuclear import and cell-cycle progression by mutated forms of the dynamin-like GTPase MxB. Proc. Natl Acad. Sci. USA 101, 89578962 (2004). Goujon, C. & Malim, M. H. Characterization of the alpha interferon-induced postentry block to HIV-1 infection in primary human macrophages and T cells. J. Virol. 84, 92549266 (2010). Yamashita, M., Perez, O., Hope, T. J. & Emerman, M. Evidence for direct involvement of the capsid protein in HIV infection of nondividing cells. PLoS Pathog. 3, e156 (2007). Yamashita, M. & Emerman, M. Capsid is a dominant determinant of retrovirus infectivity in nondividing cells. J. Virol. 78, 56705678 (2004). Dismuke, D. J. & Aiken, C. Evidence for a functional link between uncoating of the human immunodeficiency virus type 1 core and nuclear import of the viral preintegration complex. J. Virol. 80, 37123720 (2006). Lee, K. et al. Flexible use of nuclear import pathways by HIV-1. Cell Host Microbe 7, 221233 (2010). Schaller, T. et al. HIV-1 capsid-cyclophilin interactions determine nuclear import pathway, integration targeting and replication efficiency. PLoS Pathog. 7, e1002439 (2011). Koh, Y. et al. Differential effects of human immunodeficiency virus type 1 capsid and cellular factors nucleoporin 153 and LEDGF/p75 on the efficiency and specificity of viral DNA integration. J. Virol. 87, 648658 (2013). Rihn, S. J. et al. Extreme genetic fragility of the HIV-1 capsid. PLoS Pathog. 9, e1003461 (2013). Sokolskaja, E., Sayah, D. M. & Luban, J. Target cell cyclophilin A modulates human immunodeficiency virus type 1 infectivity. J. Virol. 78, 1280012808 (2004). Stoye, J. P. Fv1, the mouse retrovirus resistance gene. Rev. Sci. Tech. 17, 269277 (1998). Yamashita, M. & Emerman, M. Cellular restriction targeting viral capsids perturbs human immunodeficiency virus type 1 infection of nondividing cells. J. Virol. 83, 98359843 (2009). De Iaco, A. et al. TNPO3 protects HIV-1 replication from CPSF6-mediated capsid stabilization in the host cell cytoplasm. Retrovirology 10, 20 (2013).

METHODS SUMMARY
Gene expression in monocytoid cell lines was measured using human HT12 Expression Beadchip (Illumina) containing ,48,000 transcript probes, according to the manufacturers instructions. Candidate antiviral genes, MX2 and MX2 mutants were expressed in K562, HOS or GHOST cells using the HIV-1-based vectors SCRPSY (which encodes TagRFP and puromycin resistance) or CSIB (which confers blasticidin resistance). MX2- and control-vector-expressing cells were used as populations or as single-cell clones in infection assays to evaluate MX2 antiviral activity. All single-cycle GFP reporter viruses were pseudotyped with VSV-G. Virus stocks were generated by transfecting 293T cells with Env-defective proviral DNA that encoded GFP in place of the nef gene, or in the case of primary HIV-1 strains, full-length proviral plasmids. Alternatively, packageable GFP-expressing retroviral vector and Gag-Pol packaging plasmids were cotransfected. Target cells in microwell plates were challenged with various doses of virus and single-cycle replication evaluated after 2 days. The proportion of cells infected with GFP reporter viruses, or replication-competent virus infection in GHOST cells (which contain an LTR-GFP indicator gene) in single cycle or spreading replication assays was measured by flow cytometry. MX2 expression was reduced in target cells using a modified lentiviral shRNA expression vector (Origene). Non-dividing target cells were generated by aphidicolin treatment for 24 h before and during infection. The abundance of viral DNA species was measured using quantitative PCR with primers directed to the GFP reporter gene, or to viral LTR sequences that are proximate only in 2-LTR circles. Western blotting was done using fluorescent antibodies and signals quantitated with a LI-COR Odyssey scanner. Deconvolution microscopy and image analysis was done using a Deltavision microcopy suite.

Acknowledgements We thank members of The Rockefeller University Genomics Resource Center for assistance with the microarray experiments and members of the Bieniasz laboratory for discussion and advice. This work was supported by grants from the National Institutes of Health; R37AI64003 (to P.D.B.), R01AI078788 (to T.H.) R01AI100720 (to M.Y.), AI091707 to C.M.R., AI057158 (to I. Lipkin, Northeast Biodefense Center, subcontracted to C.M.R.) and DK095031 to J.W.S., the Greenberg Medical Research Institute and the Starr Foundation (C.M.R.) and by the Howard Hughes Medical Institute. Author Contributions M.K., S.S.Y., J.B., S.B.K., T.Z. and S.J.W. designed and executed the experiments and analysed the data. J.W.S. and C.M.R. provided an interferon-stimulated gene library and advice. M.Y. provided reagents and advice. T.H. provided reagents and advice and supervised the work. P.D.B. conceived the study, supervised the work and wrote the paper, with additional input from all authors. Author Information Reprints and permissions information is available at www.nature.com/reprints. The authors declare no competing financial interests. Readers are welcome to comment on the online version of the paper. Correspondence and requests for materials should be addressed to P.D.B. (pbienias@adarc.org).

4. 5.

5 6 6 | N AT U R E | VO L 5 0 2 | 2 4 O C T O B E R 2 0 1 3

2013 Macmillan Publishers Limited. All rights reserved

LETTER RESEARCH
METHODS
Microarray analyses. Total RNA was extracted, using the RNeasy Plus Mini kit (Qiagen), from THP-1, K562 and U937 cells that were either undifferentiated, or differentiated using PMA, and untreated or treated with 1,000 U ml21 IFN-a for 24 h before collection. Complementary RNA was prepared and probed using Human HT12 Expression Beadchip (Illumina), containing ,48,000 transcript probes, according to the manufacturers instructions. Plasmid construction. The HIV-1-based expression vector SCRPSY expresses a red fluorescent protein (TagRFP) and puromycin resistance from a TagRFPFMDV2A-Puro cassette in the unspliced viral transcript (in place of Gag coding sequence). SCRPSY also expresses spliced transcripts that encode HIV-1 Tat and Rev proteins, as well as genes inserted in Gateway-compatible sequences in place of Nef-coding sequences. The HIV-1-based expression vectors CSIB, HA-CSIB and CSIB-HA are derived from CSGW24 by replacing GFP coding sequences with a multi-cloning site (with or without sequences encoding a HA tag) followed by an internal ribosome entry site sequence and a blasticidin-resistance cassette. Candidate antiviral genes, including MX2, were also transferred into the SCRPSY vector from a library of IFN-stimulated genes8 in Gateway-compatible entry vectors using the clonase reaction (Life Technologies). Alternatively, open reading frames were amplified from complementary DNA prepared from THP-1 cells following treatment with 1,000 U ml21 IFN-a for 24 h. The GTP-hydrolysisdefective MX2 mutant (T151A), the GTP-binding-defective MX2 mutant (K131A) and MX2 lacking the N-terminal NLS (delN25)10,11 were generated using PCRbased site-directed mutagenesis. For insertion into CSIB-based vectors, MX1 and wildtype and mutant MX2 genes were amplified by PCR and inserted following SfiI restriction digestion. Cells. The adherent cell lines 293T and HOS were maintained in DMEM. The suspension cell lines U937 and K562 were maintained in RPMI. THP-1 cells were maintained in RPMI supplemented with 0.05 mM beta-mercaptoethanol. These cell lines were obtained from the ATCC. Single-cell clones of GHOST cells expressing CXCR4 or CCR5 obtained through the AIDS Reagent Program (Division of AIDS, NIAID, NIH from V. N. KewalRamani and D. R. Littman) were derived by limiting dilution and maintained in DMEM supplemented with 2.5 mg ml21 puromycin, 50 mg ml21 hygromycin and 500 mg ml21 G418. All growth media were supplemented with 10% FCS and gentamicin. Derivatives of K562 and GHOST cells expressing wild-type and mutant MX2 proteins were generated by transduction with CSIB-based vectors followed by selection in 5 mg ml21 blasticidin. HOS cells expressing wild-type or mutant MX2 proteins were prepared by transduction with SCRPSY-based vectors followed by selection in 2.5 mg ml21 puromycin. HOS cells expressing wild-type or mutant HA-tagged MX1 or MX2 proteins were prepared by transduction with HA-CSIB based vectors followed by selection in 5 mg ml21 blasticidin. In general, the MX2expressing cells were used as populations of blasticidin- or puromycin-resistant cells, but some experiments used single-cell clones of MX2-expressing HOS and K562 cells, which were derived from puromycin- or blasticidin-resistant populations by limiting dilution. For experiments in which the panels of candidate antiviral genes were screened (Fig. 2a), K562 cells were transduced with SCRPSY-based vectors, but were not selected in puromycin; infection was measured in the TagRFP-positive population. To generate the MX2-expressing or candidate gene cell lines, SCRPSY or CSIB vector stocks for transduction were generated by co-transfection of 293T cells with a VSV-G expression plasmid, an HIV-1NL4-3 Gag-Pol expression plasmid, and an MX2-expressing vector using polyethyleneimine (PolySciences). Primary CD41 T cells were isolated from human blood by Ficoll-Paque gradient centrifugation and negative selection (RosetteSep Human CD41 T Cells Enrichment Cocktail, StemCell Technologies). Cells were activated with phytohaemagglutinin (Sigma, 5 mg ml21) or anti-CD3/CD28 beads (Dynabeads Human T-Activator CD3/CD28, Gibco) for 48 h, cultured in the presence of interleukin-2 (50 U ml21, PeproTech) and treated with or without IFN-a for 24 h. Primary macrophages were differentiated from fresh human peripheral blood mononuclear cells. Cells were plated in serum-free medium for 3 h at 37 uC, the supernatant with non-adherent cells was discarded and adherent monocytes were cultured in RPMI with 10% FCS, 1% L-glutamine and granulocytemacrophage colony-stimulating factor (GM-CSF) (100 ng ml21, PeproTech) for 6 days and treated with or without IFN-a for 24 h. Viruses. All viruses were generated by transfection in 293T cells using polyethyleneimine (PolySciences). For the GFP reporter proviral plasmidsHIV-1NL4-3 DEnv-GFP (wild type, G89V, A92E and G94D CA mutants)25, NHGCapNM (wild type and N57S CA mutant)19, pLai3DEnvGFP (wild type and N74D CA mutant)14, SIVMAC DEnv-GFP, SIVAGMTan DEnv-GFP, SIVAGMSab DEnv-GFP, HIV-2ROD DEnv-GFP2610 mg of proviral plasmid was co-transfected with 1 mg of VSV-G expression plasmid. For HIV-1 (in Figs 1 and 2a), MLV, EIAV and FIV, threeplasmid vector systems were also used to generate GFP reporter viruses, whereby 5 mg of Gag-Pol, 5 mg of packageable genome and 1 mg of VSV-G expression plasmids27,28 were co-transfected. Plasmids containing full-length replicationcompetent transmitted founder HIV-1 proviruses (pTRJO.c/2851 and pREJO.c/ 2864)29 were obtained through the AIDS Reagent Program (Division of AIDS, NIAID, NIH from J. Kappes and C. Ochsenbauer). At 48 h after transfection, viral supernatants were collected and their infectivity was determined using MT4 or GHOST target cells. Infection assays. Single-cycle infectivity in HOS, K562, THP-1 and U937 cell lines was measured in cells seeded in 96-well plates at 5 3 103 cells per well and inoculated with serial-dilutions of VSV-G-pseudotyped GFP reporter viruses in the presence of 5 mg ml21 polybrene. For replication-competent primary HIV-1 strains, GHOST-R5 cells, which contain an HIV-2-LTR-GFP reporter construct, were used. Two days post-infection, cells were trypsinized (adherent cells only) and fixed in 2% paraformaldehyde. In some experiments, cells were pre-treated with IFN-a (Sigma-Aldrich) for 24 h, re-plated, followed by inoculation with GFP reporter viruses. For experiments in which infection of dividing and non-dividing cells was compared, single-cell clones of K562 and HOS cells transduced with MX2-expressing or empty vector (CSIB in the case of K562, SCRPSY in the case of HOS) were seeded at 3 3 104 cells per well in 48-well plates. Cells were treated with aphidicolin (1 mg ml21, Sigma-Aldrich) or an equivalent volume of dimethylsulphoxide alone for 24 h before infection with HIV-1 or MLV GFP reporter viruses. For spreading replication assays, empty vector or MX2-expressing HIV-1 (CSIB)-transduced GHOST-X4 cell lines (which contain an HIV-2-LTR-GFP indicator gene) were seeded at 1 3 105 cells in a six-well plate. Thereafter they were inoculated with HIV-1NL4-3 at a m.o.i. of 0.01. Cells were split at a 1:4 dilution, and the percentage of infected (GFP-expressing) cells measured, every 2 days until the empty vector cultures died owing to HIV-1-induced cytopathic effects. In all infection assays, infected cells (% GFP positive of viable cells) were enumerated by FACS analysis using a CyFlow cytometer (Partec) coupled to a Hypercyte Autosampler (Intellicyt). For cells transduced with the SCRPSY vector, the percentage of infected cells was determined as the percentage of RFP/GFP double-positive cells in the total RFP-positive population. RNA interference. Lentiviral MX2-specific shRNA (target sequence: 59-TCG CTATTCCTGGCTGCTTCAAGAGCAGA-39) or scrambled negative control shRNA pGFP-C-shLenti expression plasmids (OriGene) were digested with Xba1 and Bsu361 to excise the cytomegalovirus (CMV) promotor and GFP. Virus stocks for transducing the modified shRNA expression vectors were generated by co-transfection of 293T cells with VSV-G and HIV-1NL4-3 Gag-Pol expression plasmids. THP-1 and HOS cells were transduced and selected in 1 mg ml21 and 2.5 mg ml21 puromycin, respectively. Puromycin-selected cells were treated with IFN-a for 24 h before assessment of MX2 expression by western blotting and inoculation with GFP reporter viruses. Measurement of HIV-1 DNA species in infected cells. For analysis of HIV-1 reverse transcription products and 2-LTR circles in infected cells, 1 3 105 HOS or K562 cells were seeded in 24-well culture plates and infected with a VSV-Gpseudotyped HIV-1 reporter virus (CCGW), which is derived from CSGW but encodes a GFP reporter under the control of a CMV promoter. The virus inoculum was pretreated with 20 U ml21 RNase-free DNase I (Roche) for 1 h at 37 uC in the presence of 6 mM MgCl2. In some experiments, cells were pretreated with IFN-a for 24 h, or were treated with nevirapine starting at the time of infection. Cells were collected at 2, 12 and 24 h post-infection, washed with 13 PBS, and total DNA was extracted using the QIAamp DNA Blood Mini Kit (Qiagen). The resulting DNA samples were used as template for quantitative PCR using FastStart Universal SYBR Green Master Mix (Roche) and ABI 7500 Fast PCR system. The primer pairs used in this study are as follows: for GFP (late reverse transcription products): forward: 59-AAGTTCATCTGCACCACCGGCAA-39, reverse: 59-TGCACGCC GTAGGTCAGG-39; for 2-LTR circles: forward: 59-GACTCTGGTAACTAGA GATCCCTC-39, reverse: 59-TTGGGAGTGAATTAGCCCTTCCA-39. Western blotting. Cell suspensions were normalized for cell number, lysed in SDS sample buffer, separated by electrophoresis on NuPage 412% Bis-Tris gels (Novex) and blotted onto nitrocellulose membranes (GE Healthcare). Membranes were incubated with rabbit anti-MX2 (Novus Biologicals) and mouse anti-tubulin (Sigma) antibodies, followed by incubation with goat anti-rabbit IRDye 800CW and goat anti-mouse IRDye 680RD, respectively (LI-COR Biosciences). A LI-COR Odyssey scanner was used to detect and quantify fluorescent signals. Microscopy. HOS cells expressing N-terminally HA-tagged MX2 and MX1 were seeded onto 24-well gelatin-coated glass-bottomed dishes (MatTek) and immunostained using a combination of anti-HA (Covance) and anti-NUP98 (Cell Signaling Technology) antibodies followed by goat anti-mouse Alexa 488 and goat anti-rabbit Alexa 594 secondary antibodies (Molecular Probes). Cells were visualized by deconvolution microscopy as described previously30. Image generation and co-localization analysis and were completed with the SoftWorx software suite

2013 Macmillan Publishers Limited. All rights reserved

RESEARCH LETTER
(Applied Precision). Pearsons coefficient values were derived for MX2 or MX1 (as a control) and NUP98 by analysis of optical sections coincident with the dorsal nuclear surface, for 610 individual cells.
24. Bainbridge, J. W. et al. In vivo gene transfer to the mouse eye using an HIV-based lentiviral vector; efficient long-term transduction of corneal endothelium and retinal pigment epithelium. Gene Ther. 8, 16651668 (2001). Hatziioannou, T., Cowan, S., Von Schwedler, U. K., Sundquist, W. I. & Bieniasz, P. D. Species-specific tropism determinants in the human immunodeficiency virus type 1 capsid. J. Virol. 78, 60056012 (2004). 26. 27. 28. 29. Hatziioannou, T., Cowan, S., Goff, S. P., Bieniasz, P. D. & Towers, G. J. Restriction of multiple divergent retroviruses by Lv1 and Ref1. EMBO J. 22, 385394 (2003). Mitrophanous, K. et al. Stable gene transfer to the nervous system using a non-primate lentiviral vector. Gene Ther. 6, 18081818 (1999). Kemler, I., Barraza, R. & Poeschla, E. M. Mapping the encapsidation determinants of feline immunodeficiency virus. J. Virol. 76, 1188911903 (2002). Ochsenbauer, C. et al. Generation of transmitted/founder HIV-1 infectious molecular clones and characterization of their replication capacity in CD4 T lymphocytes and monocyte-derived macrophages. J. Virol. 86, 27152728 (2012). Jouvenet, N. et al. Plasma membrane is the site of productive HIV-1 particle assembly. PLoS Biol. 4, e435 (2006).

25.

30.

2013 Macmillan Publishers Limited. All rights reserved

LETTER RESEARCH

Extended Data Figure 1 | Candidate anti-HIV-1 genes from the microarray analysis. mRNA levels, determined using Illumina BeadChips and given in arbitrary units, for genes whose differential induction in undifferentiated and

PMA-treated THP-1, K562 and U937 cells correlated best with the anti-HIV-1 effect of IFN-a.

2013 Macmillan Publishers Limited. All rights reserved

RESEARCH LETTER

Extended Data Figure 2 | Additional candidate anti-HIV-1 genes from the microarray analysis. mRNA levels, determined using Illumina BeadChips and given in arbitrary units, for genes whose differential induction in

undifferentiated and MA-treated THP-1, K562 and U937 cells correlated to some degree with the anti-HIV-1 effect of IFN-a.

2013 Macmillan Publishers Limited. All rights reserved

LETTER RESEARCH

Extended Data Figure 3 | Induction of MX2 by IFN-a in primary CD41 T cells and macrophages. Western blot analysis of MX2 and tubulin expression in purified CD41 T cells, activated with PHA or anti-CD3/CD28, and macrophages treated for 24 h with the indicated doses of IFN-a. Numbers below each lane indicate fluorescence intensity associated with the MX2 band. The second more rapidly migrating MX2 species that was detected inconsistently is of unknown provenance, and may represent a proteolytic breakdown product, or may arise through the use of an alternative start codon at amino acid 25, generating an MX2 protein that lacks the NLS.

2013 Macmillan Publishers Limited. All rights reserved

RESEARCH LETTER

Extended Data Figure 4 | MX2 is required for the full antiviral activity of IFN-a in HOS cells. a, Western blot analysis of MX2 expression HOS cells transduced with vectors expressing control or MX2-targeted shRNAs, and treated with IFN-a. Numbers below each lane indicate fluorescence intensity associated with the MX2 band. b, Infectious titre of an HIV-1GFP reporter virus determined using the shRNA-containing HOS cells from a, with or without IFN-a treatment. Titres are mean 1 s.d., n 5 3 technical replicates, P values calculated using unpaired t-test, representative of three experiments.

2013 Macmillan Publishers Limited. All rights reserved

LETTER RESEARCH

Extended Data Figure 5 | MX2 activity reduces levels of nuclear HIV-1 DNA in K562 cells. Quantitative PCR analysis of reverse transcript (left) and 2-LTR circle (right) abundance in empty-vector untreated (none) nevirapine-treated or MX2-expressing K562 cells.

2013 Macmillan Publishers Limited. All rights reserved

RESEARCH LETTER

Extended Data Figure 6 | Localization of MX2 at nuclear pores. a, Deconvolution microscopic images (single optical sections) of immunofluorescently stained NUP98 (red), haemagglutinin (HA)-tagged MX2 (green, expressed using CSIB vectors) and DAPI (49,6-diamidino-2phenylindole)-stained DNA (blue) in HOS cells. The top set of panels is an optical section approximately through the centre of the vertical dimension of the nucleus, whereas the middle and bottom panels are an optical section

approximately coincident with the dorsal surface of the nucleus. The bottom panels are an expanded view of a portion of the centre panels. Scale bars, 10 mm (top), 5 mm (middle) and 1 mm (bottom). b, Pearsons coefficient for colocalization of MX1 or MX2 and NUP98. Each data point represents an individual cell and the horizontal bar is the mean (n 5 6 for MX1, n 5 10 for MX2).

2013 Macmillan Publishers Limited. All rights reserved

LETTER RESEARCH

Extended Data Figure 7 | The N57S capsid mutation reduces HIV-1 sensitivity to IFN-a in HOS cells. Infectivity of wild-type and N57S CA-mutant HIV-1GFP reporter viruses in untreated and IFN-a-treated HOS cells. Titres are mean 1 s.d, n 5 3 technical replicates, representative of three experiments. Fold inhibition is the ratio of the titres on untreated and IFN-a-treated cells.

2013 Macmillan Publishers Limited. All rights reserved

RESEARCH LETTER

Extended Data Figure 8 | Effect of MX2 on HIV-1 and murine leukaemia virus infection in dividing and non-dividing cells. a, MLVGFP reporter virus infection of dividing and non-dividing (aphidicolin-treated) vector or MX2-expressing HOS cell clones. b, HIV-1GFP reporter virus infection of dividing and non-dividing (aphidicolin-treated) vector or MX2-expressing K562 cell clones. c, MLVGFP reporter virus infection of dividing and non-dividing (aphidicolin-treated) vector or MX2-expressing K562 cell clones.

2013 Macmillan Publishers Limited. All rights reserved